Rickettsia rickettsii virulence determinants RARP2 and RapL mitigate IFN-β signaling in primary human dermal microvascular endothelial cells

ABSTRACT We compared the growth characteristics of a virulent Rickettsia rickettsii strain (Sheila Smith) to an attenuated R. rickettsii stain (Iowa) and a non-pathogenic species (R. montanensis) in primary human dermal microvascular endothelial cells (HDMEC). All replicated in Vero cells, however, only the Sheila Smith strain productively replicated in HDMECs. The Iowa strain showed minimal replication over a 24-h period, while R. montanensis lost viability and induced lysis of the HDMECs via a rapid programmed cell death response. Both the virulent and attenuated R. rickettsii strains, but not R. montanensis, induced an interferon-1 response, although the response was of lesser magnitude and delayed in the Sheila Smith strain. IFN-β secretion correlated with increased host cell lysis, and treatment with anti-IFNAR2 antibody decreased lysis from Iowa-infected but not Sheila Smith-infected cells. Both Sheila Smith- and Iowa-infected cells eventually lysed, although the response from Sheila Smith was delayed and showed characteristics of apoptosis. We, therefore, examined whether reconstitution of the Iowa strain with two recently described putative virulence determinants might enhance survival of Iowa within HDMECs. Reconstitution with RARP2, which is inhibitory to anterograde trafficking through the Golgi apparatus, reduced IFN-β secretion but had no effect on cell lysis. RapL, which proteolytically processes surface exposed autotransporters and enhances replication of Iowa in Guinea pigs, suppressed both IFN-β production and host cell lysis. These findings suggest distinct mechanisms by which virulent spotted fever group rickettsiae may enhance intracellular survival and replication. IMPORTANCE We examined a naturally occurring non-pathogenic rickettsial species, R. montanensis, a laboratory-attenuated R. rickettsii strain (Iowa), and a fully virulent R. rickettsii strain (Sheila Smith) for growth in human dermal microvascular endothelial cells. The two avirulent strains replicated poorly or not at all. Only the virulent Sheila Smith strain replicated. IFN-β production correlated with the inhibition of R. rickettsii Iowa. Reconstitution of Iowa with either of two recently described putative virulence determinants altered the IFN-β response. A rickettsial ankyrin repeat protein, RARP2, disrupts the trans-Golgi network and inhibits IFN-β secretion. An autotransporter peptidase, RapL, restores proteolytic maturation of outer membrane autotransporters and diminishes the IFN-β response to enhance cell survival and permit replication of the recombinant strain. These studies point the way toward discovery of mechanisms for innate immune response avoidance by virulent rickettsia.

ancestral group (1).R. rickettsii, the etiologic agent of Rocky Mountain spotted fever (RMSF), causes the most severe disease among the SFG rickettsiae.Even within the species R. rickettsii, there is great variation in virulence (2)(3)(4)(5).This disparity in severity of disease was noted very early in the history of Rocky Mountain spotted fever when mortality from RMSF was over 80% in the Bitterroot Valley of western Montana but only about 5% in nearby Idaho (6).The reasons behind these strain-dependent differences in virulence are only now being deciphered.Many rickettsial species are capable of infecting humans or other mammals to cause disease of varying degrees of severity, while other species have never been associated with human disease or shown to cause disease in experimental animal models.Certain of these non-pathogenic rickettsiae may simply be endosymbionts or commensals of arthropods incapable of mammalian infections (7).Efforts to examine rickettsial pathogenesis have been limited by the paucity of genetic tools and the lack of verified in vitro systems that model aspects of rickettsial pathogenesis.Deeper insights into early interactions between human host cells and virulent and avirulent rickettsial strains and species are required to understand the inherent susceptibility or resistance of humans to rickettsial infections.
Rickettsiae can infect several cell types but have tropisms for endothelial cells and are initially observed within the microvasculature following transmission of rickettsia via tick bite (8)(9)(10)(11).Most symptoms from rickettsial infections can be attributed to "rickettsial vasculitis"; disseminated infection and inflammation of vascular tissue that diminishes vascular functions (12).Many of the symptoms of RMSF, including the characteristic rash, can be attributed to a breakdown of vascular permeability.Severe RMSF disease can result in widespread necrosis, organ failure, gangrene, and death.In addition to their barrier and vascular functions, endothelial cells are active contribu tors to innate immunity; they can detect pathogens, express antimicrobial genes, and recruit professional immune cells to sites of infection (13).Thus, subverting innate immune signaling within their initial microvascular endothelial host cells is an early, but underappreciated, aspect of rickettsial pathogenesis.Primary endothelial cells have been used as in vitro model to study virulent rickettsiae (14).However, comparative infections between virulent and avirulent rickettsiae are in need of further examination.
Here, we describe that infection of primary human dermal microvascular endothelial cells (HDMECs) with rickettsiae leads to different outcomes depending on the patho genic potential of the rickettsiae.Specifically, HDMECs infected with non-pathogenic Rickettsia montanensis rapidly lyse to deprive the bacteria of its essential cytosolic niche.HDMECs infected with the attenuated R. rickettsii strain Iowa survive somewhat longer and initiate a potent type-I interferon (IFN-I) response that induces host cell death and inhibits bacterial growth.HDMECs infected with the highly virulent R. rickettsii strain Sheila Smith survive the longest, and their IFN-I responses are delayed and minimized such that they poorly express antimicrobial proteins and support robust bacterial growth before dying by apoptosis.Thus, primary human endothelial cells appear to have multiple layers of innate immune defenses that restrict non-pathogenic rickettsiae, but virulent rickettsiae have evolved mechanisms to successfully inhibit early host cell death and preserve the intracellular niche that is essential to the life cycles of these microbes.
In an effort to define virulence factors of R. rickettsii Sheila Smith that may contrib ute to an abated IFN-β response, promote host cell survival, and enhance rickettsial replication, we complemented the Iowa strain with two recently described putative virulence factors, rickettsial ankyrin repeat protein 2 (RARP2) (15,16) and RapL (17).RARP2 is an ER-localized cysteine protease that disrupts the trans-Golgi network (TGN) to inhibit trafficking to the host plasma membrane (15,16).RapL is an autotransporter protease that cleaves the passenger domain from the autotransporter domain of spotted fever group autotransporters (17).Each individually moderated the IFN-β response by different mechanisms.RARP2 reduced secretion of IFN-β into the medium, while RapL induced a reduction in total IFN-β synthesis and corresponding increase in cell survival and rickettsial replication.

Primary human dermal microvascular endothelial cells restrict avirulent rickettsiae
Vero cells are immortalized epithelial cells deficient in IFN-I signaling (18).This may contribute to their utility as permissive hosts for intracellular pathogens, including diverse rickettsiae (18,19).Vero cells were infected with R. montanensis, R. rickettsii Iowa, and R. rickettsii Sheila Smith, and numbers of viable rickettsiae enumerated at 2 and 24 h post-infection.Non-pathogenic R. montanensis and R. rickettsii Iowa replicated approxi mately 15-fold in Vero cells over 24 h, and virulent R. rickettsii Sheila Smith increased approximately 30-fold (Fig. 1A).To assess survival of the Vero cell monolayer, culture supernatants were assayed for activity of the cytosolic enzyme, lactate dehydrogenase (LDH), over 24 h of infection.Infections of Vero cells did not lead to substantial increases of LDH activity (Fig. 1B).Thus, Vero cells allow pathogenic and non-virulent rickettsiae to grow to high titers, and, despite their large bacterial burdens, the host cells do not lyse over this time course.
To determine if a potentially more relevant cell type might respond differently, a cell culture model of primary human dermal microvascular endothelial cells was adopted, and their interactions with rickettsia were assessed.R. montanensis induced lysis of infected cells as early as 6 h post infection and showed decreasing numbers of viable rickettsiae (Fig. 1C and D).R. rickettsii Iowa populations showed a minimal increase in number but maintained viability for up to 24 h in culture and showed elevated LDH release by 12 h post-infection.In contrast, R. rickettsii Sheila Smith populations replica ted 10-fold (Fig. 1C) and delayed onset of cell lysis until 24 h post-infection (Fig. 1D).Uninfected HDMECs or those treated with heat-killed rickettsiae did not show apprecia ble lysis.These data indicate that primary human endothelial cells are able to restrict non-pathogenic Rickettsia spp.growth, likely by activating mechanisms of programmed cell death to eliminate the intracellular niche.However, virulent R. rickettsii appears to delay the onset of cell death and is thereby able to replicate within these host cells.As such, HDMECs provide a useful in vitro model to study rickettsial pathogenesis.
To establish whether non-pathogenic and pathogenic rickettsiae differentially induce apoptosis in HDMECs, Caspase-3/7 activities were assessed, and cleavage of caspase-3 target proteins, poly ADP-ribose polymerase (PARP) and gasdermin E (GSDME), were examined (20,21).R. rickettsii Sheila Smith infections increased Caspase-3/7 activities (Fig. 2A) and induced cleavage of PARP and GSDME (Fig. 2B).R. rickettsii Iowa infections modestly increased Caspase-3/7 activities, but cleavage of PARP and GSDME was not observed after R. montanensis or R. rickettsii Iowa infections.These data demonstrate that virulent R. rickettsii induce late-onset apoptosis and infer that non-virulent rickettsiae induce non-apoptotic mechanisms of cell death within HDMEC hosts at earlier time points, likely contributing to the poor replication of the avirulent rickettsiae.

R. rickettsii Sheila Smith minimizes and delays IFN-β secretion, STAT1/2 phosphorylation, and ISG expression
IFN-I are potent proinflammatory cytokines whose autocrine and paracrine signaling can activate antimicrobial gene expression or host cell death to restrict infections by intracellular microbes (22), including rickettsia (23)(24)(25).IFN-β was quantified in culture supernatants from HDMECs infected with rickettsiae for 24 h.HDMECs infected with R. rickettsii Iowa or Sheila Smith secreted substantial amounts of IFN-β into culture supernatants, whereas uninfected controls or those treated with heat-killed rickettsiae or viable R. montanensis did not (Fig. 3A).Levels of secreted IFN-β after R. rickettsii Sheila Smith infections were reduced as compared to R. rickettsii Iowa.To examine whether there were kinetic differences in the secretion of IFN-β after R. rickettsii infections, supernatants were assayed after 6, 12, 18, and 24 h post infection (HPI) (Fig. 3B).IFN-β was detectable as early as 6 HPI when HDMECs were infected with R. rickettsii Iowa, whereas the cytokine was first detected in Sheila Smith-infected cultures at 12 HPI.At all time points examined, IFN-β levels were substantially higher in supernatants from R. rickettsii Iowa-infected cells compared to Sheila Smith infections.These data suggest that IFN-I production and signaling may be distinct in HDMECs infected with R. rickettsii strains of different virulence potentials.The data further suggest that R. rickettsii Sheila Smith infections minimize and delay secretion of IFN-β relative to the Iowa strain.IFN-β signals through the IFN-α receptors (IFNAR1 and IFNAR2) lead to phosphoryla tion of signal transducer and activator of transcription 1 and 2 (STAT1 and STAT2) transcription factors, which activate expression of interferon-stimulated genes (ISGs) (22).To assess whether IFN-β downstream signaling occurs in HDMECs upon R. rickettsii infections, Western blotting for STAT1 and STAT2 was performed.R. rickettsii Iowa infections induce STAT1 and STAT2 phosphorylation as early as 6 HPI, whereas R. rickettsii Sheila Smith infections exhibited STAT1 and STAT2 phosphorylation by 12 HPI (Fig. 3C).These data demonstrate that the delayed IFN-β secretion mediated by R. rickettsii Sheila Smith infections also delays downstream activation of STAT1 and STAT2 transcription factors.
ISGs comprise a large collection of genes, with diverse antimicrobial roles, that are positively regulated by IFN signaling (22,26).CXCL9, CXCL10, and CXCL11 are IFNregulated, secreted, C-X-C-type chemokines which can recruit professional immune cells to sites of inflammation (27).Additionally, these chemokines have broad antibacterial activity which may impede bacterial growth at high, local concentrations (28).CXCL10 secretion was measured by ELISA using supernatants from HDMECs infected with rickettsiae.R. rickettsii Iowa induced robust secretion of CXCL10 and released over 500fold more of this chemokine than Sheila Smith at 12 HPI (Fig. 3D).Uninfected controls, those treated with heat-killed rickettsia, or R. montanensis did not secrete CXCL10.
Expression of the ISGs 2′-5′-oligoadenylate synthetase-like (OASL), retinoic acid inducible gene I (RIG-I), interferon regulatory factor 1 (IRF-1), TNF-related apoptosisinducing ligand (TRAIL), and indoleamine 2,3-dioxygenase (IDO) was assessed by Western blotting.Despite differences in IFN-β secretion and STAT1/2 phosphorylation, HDMECs express similar amounts of OASL, RIG-I, and IRF-1 after 12 h of infection with R. rickettsii Iowa or Sheila Smith (Fig. 3E).However, R. rickettsii Iowa infections also induced IDO and TRAIL expression at 12 HPI, while Sheila Smith infections did not.By 24 HPI, Sheila Smith infections eventually induced TRAIL, but IDO expression was not detected.These data imply that in addition to minimizing and delaying IFN-β secretion, R. rickettsii Sheila Smith may have additional mechanisms to repress expression of a subset of ISGs which may mediate host cell death (TRAIL) or have antibacterial activities (CXCL10 and IDO).TRAIL is a secreted protein that can either promote inflammation or activate different mechanisms of cell death (29).IDO is a tryptophan-degrading enzyme that starves intracellular microbes of this amino acid to limit intracellular growth (30,31).The data could also suggest that the IFN-I response initiated by R. rickettsii Iowa participates in the restriction of bacterial growth in HDMECs, while the delayed response induced by Sheila Smith does not control these virulent rickettsia.

Blocking IFNAR2 signaling rescues R. rickettsii Iowa growth in HDMECs
The robust IFN-I response induced by R. rickettsii Iowa infections of HDMECs may restrict growth of these bacteria by promoting premature host cell death and expression of antimicrobial genes.This was tested by utilizing a blocking antibody that binds to IFNAR2 and reduces IFN-I signaling.Anti-IFNAR2 treatment reduced host cell death after infections of R. rickettsii Iowa but not Sheila Smith (Fig. 4A).Importantly, treatment with anti-IFNAR2 enabled R. rickettsii Iowa intracellular replication in HDMECs, presumably by enhancing cell survival, while treatment with this antibody did not affect the growth of Sheila Smith (Fig. 4B).Treatment of HDMECs with anti-IFNAR2 did not reduce phosphory lation of STAT1 and STAT2 after infections with R. rickettsii Iowa as compared to treat ment with isotype matched, negative control antibodies (Fig. 4C).Curiously, anti-IFNAR2 was more effective at reducing STAT1/2 phosphorylation after Sheila Smith infections, suggesting that other ligand and receptor platforms may be signaling through STAT1 and STAT2 upon Iowa infections.Nonetheless, anti-IFNAR2 treatment was sufficient to reduce expression of TRAIL and IDO after R. rickettsii Iowa infections at the later time points (Fig. 4D).These data demonstrate that the IFN-I signaling induced by HDMECs infected with R. rickettsii Iowa results in proinflammatory and antimicrobial gene expression which untimely causes host cell death and restricts bacterial replication.The data also show that the reduced IFN-I signaling induced by Sheila Smith infections does not result in IDO expression and premature host cell death, nor limit bacterial growth.IFN-I signaling is thus likely a key aspect of its pathogenesis in the vasculature of infected mammals.Recognizing the virulence alleles that enable Sheila Smith to subvert IFN-I signaling is important to understanding rickettsial pathogenesis.

Complementation of R. rickettsii Iowa with Sheila Smith rapL promotes HDMEC survival and enables intracellular replication
The genomes of R. rickettsii Iowa and Sheila Smith are over 99% identical (3), although virulence was lost from the Iowa strain during extensive laboratory passage (32).We have documented genetic differences between these R. rickettsii strains (2, 3) and have investigated the contribution of different alleles to Rickettsia virulence (2, 15-17, 33, 34).
Two of these suspected or proven virulence determinants were selected for further investigation.RARP2 inhibits protein secretion from infected cells (15) and could potentially interfere with IFN-Ι secretion.RapL proteolytically cleaves outer membranelocalized autotransporters to free the surface-exposed passenger domains from the membrane-embedded autotransporter domains and create potentially soluble cytosolic products.RapL enhances replication of recombinant Iowa strains in animal model systems and restores a modest degree of virulence in a Guinea pig model of infection by unknown mechanisms (17).
We investigated whether HDMECs infected with R. rickettsii Iowa strains complemen ted with Sheila Smith rarP2 or rapL alleles might have altered host-pathogen interactions.LDH activity was determined in HDMEC supernatants over 24 h of infection with R. rickettsii Iowa, Sheila Smith, or Iowa strains carrying different plasmids.HDMECs infected with Iowa complemented with Sheila Smith rapL survived dramatically longer than those complemented with the pRAM-FC1 parent vector, comparable to Sheila Smith infec tions (Fig. 5A).Sheila Smith infections induce a delayed and diminished IFN-I response compared to the Iowa strain.To determine if RARP2 or RapL influences this phenotype, HDMEC supernatants and monolayer lysates were assessed for IFN-β production after 12 h of infection with R. rickettsii Iowa, Sheila Smith, or Iowa strains carrying complemen tation plasmids.As shown above, supernatants from HDMECs infected with R. rickettsii Iowa had high levels of IFN-β, while those from Sheila Smith-infected HDMECs were reduced (Fig. 5B).However, lysates from monolayers of Sheila Smith-infected HDMECs retained much higher intracellular IFN-β levels than those from Iowa-infected cells.Furthermore, cells infected with Iowa complemented with Sheila Smith RARP2 secre ted reduced amounts of IFN-β, but their lysates had more IFN-β than those with the control plasmid.Only the two strains expressing Sheila Smith RARP2 showed more IFN-β in the cellular lysates than secreted into the media.These data are consistent with TGN dispersion by Sheila Smith RARP2 and inhibition of protein secretion (15).Iowa strains expressing Sheila Smith rapL also had moderately reduced IFN-β secretion, but accumulated chemokine was not found in cellular lysates.These data suggest that RapL may lower IFN-β expression as a means of promoting HDMEC viability.
Intracellular replication was assessed to determine if the enhanced HDMEC survival mediated by rapL complementation enabled R. rickettsii Iowa growth.Indeed, the Iowa strain complemented with rapL replicated 12-fold over 24 h within HDMECs, while the pRAM-FC1 parental strain only replicated 2.2-fold, a difference significant at P < 0.0001.Iowa strains expressing the rapL-S160A, which has a catalytic loss-of-function mutation, or expressing RARP2, did not grow (Fig. 5C).These data are consistent with enhanced replication of R. rickettsii Iowa complemented with RapL in a Guinea pig model (17).The data are also consistent with the inability of the RARP2-complemented Iowa strain to restore disease in laboratory animals (16).IFN-β levels were determined in HDMEC cells infected for 12 h (Fig. 5D) with the same set of complemented strains as shown in Fig. 5C.The S160A catalytically inactive RapL mutant showed increased secretion of IFN-β relative to the wild-type RapL-expressing strain, which correlates with the inability

DISCUSSION
We used primary human dermal microvascular endothelial cells as a cell culture model to study the outcomes of infections with rickettsia of different pathogenic potentials.The rickettsial strains examined included the highly virulent R. rickettsii strain Sheila Smith, the laboratory attenuated R. rickettsii strain Iowa, and the non-pathogenic R. montanensis, a commensal strain not associated with disease in humans or animals.Each resulted in distinct infection outcomes depending on pathogenic potential of the bacteria.We found that only the virulent Sheila Smith could prevent premature host cell lysis and replicate within HDMECs, while infections with non-pathogenic rickettsia resulted in premature lysis of these host cells.These interactions, which occur within a few hours after infection, provide plausible explanations for why both R. montanensis and R. rickettsii Iowa do not readily cause disease in humans or animal models.These data also indicate that microvascular endothelial cells have innate immune barriers which restrict some intracellular microbes and result in mechanisms of programmed cell death that could protect a mammalian host from microbial infection through the limited loss of compromised cells.
Spotted fever group rickettsia encompasses a spectrum of virulence, ranging from the highly virulent R. rickettsii to those such as R. montanensis that have never been associated with disease.Other SFG rickettsiae display intermediate levels of disease severity.Several studies have examined the responses of various phagocytic cell types to infection by virulent vs avirulent Rickettsia spp.(35)(36)(37)(38)(39).In THP-1 human macrophagelike cells, R. montanensis was killed and partially co-localized with lysosomal markers, LAMP1 and cathepsin D (38).The viability of the host THP-1 cells was not reported.THP-1 cells did not restrict R. conorii replication.Proteomic studies found that infections of THP-1 cells with R. montanensis and R. conorii did not activate IFN-I responses (37,38).A more recent analysis of three SFG strains of intermediate virulence (R. parkeri, R. massiliae, R. africae) induced different levels of IFN-β production and ISG expression and died by pyroptosis as indicated by the cleavage of gasdermin D (37).In mouse bone marrow-derived macrophages (BMDM), minimal restriction of R. montanensis growth was observed compared to virulent R. rickettsii and R. typhi, although R. montanensis did replicate over 24 h in BMDMs, a notable difference from our results using HDMECs.It is unclear at this point whether the failure of mouse BMDMs to restrict R. montanensis growth reflects a species-dependent difference.The R. montanensis-infected BMDMs showed increased lysis of infected cells at 24 h relative to the R. rickettsiior R. typhiinfected BMDMs.This increased lysis was accompanied by cleavage of gasdermin D, which was not observed in the R. rickettsiior R. typhi-infected BMDMs, again suggest ing pyroptosis as a mechanism of cell death in the R. montanensis-infected cells (39).A similar cleavage of gasdermin D was observed in BMDMs infected with R. parkeri (40).It is noteworthy that IFN-β was not induced by R. montanensis.It may be that R. montanensis infection leads to cellular death before an IFN-β response can be generated.The mechanism of programmed cell death in R. montanensis-infected HDMECs thus remains undefined and will be the focus of future studies.The mechanism of cell death in Iowa-infected HDMECs is also undefined but occurs later and appears to involve IFN-1 responses.The response of HDMECs to R. rickettsii infection is clearly distinct.IFN-β is a prominent response to R. rickettsii infection, although the response is moderated in Sheila Smith-infected cells relative to the high levels of secreted IFN-β in Iowa-infected cells.Suppressing the IFN-β response in Iowa-infected cells promotes cell survival and enhances rickettsial replication.
Obligate intracellular bacteria must preserve cellular functions for sufficient time to complete bacterial replication.Virulent R. rickettsii are known to inhibit apoptosis (41)(42)(43), although the precise mechanisms remain unclear.IFN-β responses are of greater magnitude and occur sooner after infection of HDMECs with attenuated R. rickettsii Iowa than that with the virulent Sheila Smith strain.Downstream effects, such as activation of STAT1 and STAT2, also occur sooner in Iowa-infected cells, as does induction of several ISGs.For example, Sheila Smith infections delayed and reduced the expression of CXCL10 and TRAIL and did not induce expression of IDO.The data suggest that these pathogenic R. rickettsii may have additional mechanisms to limit the expression of specific ISGs to further counter the deleterious consequences of IFN-I signaling.The diminished IFN-β response in Sheila Smith-infected cells is likely due to a loss of mechanisms to control IFN-β production from the highly attenuated Iowa.The Iowa strain was initially described as being of mild virulence, increasing with laboratory passage, and ultimately becoming avirulent (32).Importantly, the close relatedness of the R. rickettsii strains of Iowa and Sheila Smith (3) permitted the evaluation of specific rickettsial virulence determinants in moderation of host innate immune responses.
Rickettsial ankyrin repeat protein 2 is a type IV secreted effector protein that associates with the endoplasmic reticulum and causes dispersion of the trans-Golgi network to inhibit secretion of cellular proteins to the host plasma membrane (15,16).Iowa has an internally truncated RARP2 allele lacking 7 of the 10 ankyrin repeat units of the Sheila Smith strain RARP2 which does not localize to the ER or cause TGN disper sion.R. montanensis does not contain an RARP2 homolog (15,16).Complementation of R. rickettsii Iowa with the Sheila Smith RARP2 allele restored TGN dispersion but did not restore virulence to the Iowa strain in Guinea pigs (16).More IFN-β is retained in HDMECs infected with Iowa-expressing SS-RARP2, consistent with decreased anterog rade trafficking of cellular proteins (15).Diminished secretion of IFN-β to the extracellular environment may be beneficial to the rickettsia by reducing activation of adjacent cells and thus providing the rickettsiae with more permissive cells in which to replicate.
The mechanism of RapL inhibition of IFN-β production and cell lysis with attendantenhanced rickettsial replication is less clear.We recently demonstrated that the Iowa allele of rapL has a mutated translational start site and that complementation of Iowa with Sheila Smith rapL restored proper proteolytic processing of rickettsial autotrans porters and partially restored virulence in Guinea pigs (17).R. montanensis contains a homolog of RapL that is 96% identical to the Sheila Smith version over the full 276 amino acids of the protein.The function of RapL is proteolytic processing of surface-exposed autotransporters, of which R. rickettsii express four (rOmpA, rOmpB, Sca1, and Sca2) (44).It seems unlikely-although not impossible-that RapL functions directly upon IFN-β production but rather that the soluble passenger domain of one of the autotransporters functions to inhibit IFN-β synthesis.
Vascular endothelial cells are the primary target of rickettsial infections (11).Endothelial cells are also key players in innate immune responses to pathogenic bacteria and viruses (45).As such, they are among the first line of defense against rickettsial infections.While the mechanism(s) of programmed cell death induced by R. montanensis and R. rickettsii Iowa remain elusive, the distinct cellular responses to infection demon strate that the primary target cells of rickettsial infection possess potent anti-rickettsial activity but that virulent Rickettsia spp.have mechanisms to overcome these restrictions.Rickettsia-HDMEC interactions in cell culture offer a valuable in vitro system to investi gate rickettsial pathogenesis and mechanisms of avoidance of innate immune responses.Replication within HDMECs could also be adopted and standardized to better predict the pathogenic potential of novel Rickettsia spp.without the immediate use of laboratory animals.
For plaque determination assays, 10-fold serial dilutions of rickettsia in BHI media were added to Vero cell monolayers in 6-well plates.Following bacterial invasion, the cells were overlaid with M199 media supplemented with 5% fetal bovine serum (FBS) and 0.5% molten low-melt agarose.After 5-7 days of growth at 34°C, plaques were counterstained with thiazolyl blue tetrazolium bromide, 0.3% (MTT).The following day, plaque-forming units were counted, and PFU/mL was calculated.

Mammalian cells
Vero76 cells (Vero, ATCC CCl-81) were maintained in RPMI 1640 supplemented with 5% fetal bovine serum and were grown at 37°C.Vero cell-Rickettsia infections were performed using M199 media supplemented with 2% FBS and grown at 34°C.
Primary human dermal microvascular endothelial cells from adult or juvenile donors were purchased from PromoCell (C-12212) (Heidelberg, Germany) and maintained in Endothelial Growth Media MV 2 (PromoCell C-22121).The cells were grown at 37°C, 5% CO 2 in a humidified incubator and were maintained, stored, and split according to manufacturer's instructions.HDMEC-Rickettsia infections were performed in the same media but at 34°C.

Rickettsia replication assays
Vero or HDMECs were infected with rickettsia at an MOI of 1-2 unless otherwise stated.Bacteria were attached by centrifugation (1,136 × g, 5 min) and were allowed to invade for 30 min at 34°C, 5% CO 2 .Afterward, infection mediums were exchanged for fresh media and were incubated at 34°C, 5% CO 2 , for 2 or 24 h.Cells were scraped into BHI, disrupted using a Mini Beadbeater (BioSpec Products) for 10 seconds, and stored at −80°C until titers were determined by plaque assay (47, 48).Titers were normalized to the average PFU/mL at 2 HPI, for each strain, for each day.

LDH viability assays
Host cell lysis was assessed using the CytoTox 96 Non-Radioactive Cytotoxicity Assay (G1780) from Promega (Madison, WI) as recommended by the manufacturer.Briefly, HDMECs or Vero cells grown in 96-well plates were infected or treated with specific compounds for indicated lengths of time.Select uninfected and unstimulated samples were treated with the provided 10× Lysis Reagent prior to analysis.In a fresh plate, 50 µL of media or cell supernatants was mixed with 50 µL of reconstituted CytoTox 96 reagent, and after robust color changes were observable (approximately 5-10 min), the reaction was stopped with 50 µL of Stop Solution.The A 490 of equilibrated reactions were recorded with a Molecular Devices SpectraMax iD3 plate reader.Host cell lysis was calculated by subtracting the average A 490 of untreated media samples from each experimental sample, then normalized such that samples treated with 10× Lysis Reagent represented 100% lysis.

Western blotting
As previously described (49), cells were treated with various compounds or infected with rickettsia for various lengths of time as indicated for each experiment.Monolayers were scraped into sample buffer (50) with HALT phosphatase and protease inhibitors (Thermo Scientific, Waltham, MA), and benzonase (Sigma-Aldrich, St. Louis, MO).Samples were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto methanol-rinsed PVDF membranes with a BioRad SemiDry Transfer apparatus.Membranes were dried, methanol rinsed, and blocked with Li-Cor Intercept TBS-based blocking buffer.Membranes were probed with select antibodies diluted in blocking buffer, washed with TBST, probed with Li-Cor secondary antibodies, washed, and membranes were scanned with a Li-Cor Odyssey CLx and analyzed with Image Studio v5.2software.

Caspase-3/7 activity assays
Caspase activities were measured with Casp-Glo 3/7 Assay System from Promega corporation (Madison, WI), as directed by the manufacturer.Briefly, HDMECs were infected with Rickettsiae, treated with Staurosporine (2 µM) in 96-well white wall plates.After select times, reconstituted, 2× reagent was added to all samples, mixed well, and was allowed to equilibrate for 30 min in the dark.Luminescence was then recorded with a Molecular Devices SpectraMax iD3 plate reader, and activity was normalized to the average of untreated, uninfected samples (Ctrl).

ELISA
Measurements for the IFN-β cytokine or CXCL10 chemokine were performed with the Human IFN-beta Quantikine ELISA Kit (DIFNB0) or Human CXCL10/IP-10 Quantikine ELISA Kit (DIP100), respectively.Both assay kits were from R&D Systems (Minneapolis, MN) and were used as directed by the manufacturer.Briefly, HDMECs were infected with rickettsiae or treated with MSA-2, 50 µM for various times as indicated in each experi ment.Culture supernatants were collected, centrifuged (10,000 × g for 3 min, 4°C), and saved at −20°C until analysis.For select samples, cell monolayer lysates were generated by lysing with 0.1% Saponin in water, one freeze-thaw cycle, and were then buffered by the addition of 10× PBS.On the day of assay, standards were reconstituted and serially diluted, and experimental samples were thawed.Plates were incubated with standards or samples (in technical duplicate), washed, probed with an antibody-HRP conjugate, and developed all according to the provided protocol.A 450 were recorded, and cytokine or chemokine concentrations were calculated using linear regression.

FIG 4
FIG 4 HDMECs restrict R. rickettsii Iowa, but not Sheila Smith, through IFN-β signaling.(A) Lysis of HDMECs was determined by LDH activity in the culture supernatants after 12, 18, and 24 h of infection.Select samples were also treated with isotype control or anti-IFNAR2 (mean ± S.D., N = 6).(B) Normalized bacterial burdens within HDMECs infected with R. rickettsii Iowa or Sheila Smith treated with isotype control or anti-IFNAR2 for 24 h (mean ± S.D., N = 6).Significance in A was assessed with two-way ANOVA and one-way ANOVA in B; P < 0.05, *; P < 0.0001, ****.(C) Western blotting with anti-p-STAT1, STAT1, p-STAT2, STAT2, and GAPDH of HDMECs infected with R. rickettsii strains treated with isotype control or anti-IFNAR2 antibodies (3 mg/mL) for 6 and 12 h (representative results are shown, N = 4).(D) Western blotting with anti-TRAIL, IDO, and GAPDH of HDMECs infected with R. rickettsii strains treated with isotype control or anti-IFNAR2 for 12, 18, and 24 h (representative results are shown, N = 3), ns = not significant.

FIG 5
FIG 5 Complementation of R. rickettsii Iowa with RARP2 or RapL restores virulence phenotypes in HDMECs.(A) Lysis of HDMECs was measured over 24 h of infection with R. rickettsii strains (error bars represent mean ± S.D., N = 6).No samples were significantly different at T = 0.At 24 HPI, there was no significant difference between Iowa and Iowa + RarP2; Iowa and Iowa + FC1 were significantly different (P < 0.01); as were SS and Iowa + RapL (P < 0.01).All other samples were significantly different from each other (P < 0. 0001).Data were analyzed by one-way ANOVA using Tukey's multiple comparison adjustment.(B) ELISA measurements of IFN-β in HDMEC supernatants and monolayer lysates after 12 h of infection with R. rickettsii strains (error bars represent mean ± S.D., N = 6).Data were analyzed by two-way ANOVA with Sidak's multiple comparison test.ns = not significant; * = P < 0.05; **** = P < 0.0001.(C) Normalized rickettsial burdens within HDMECs after 2 or 24 h of infection with R. rickettsii Iowa strains complemented with control vector (FC1) or those expressing Sheila Smith RARP2, RapL, or RapL with a point mutation (RapLS160A) in the putative active site serine (17).Shown are the means ± S.D. (n = 6).Significance was assessed by two-way ANOVA with Sidak's multiple comparison test.ns = not significant; * = P < 0.05; **** = P < 0.0001.(D) IFN-β levels in supernatants from HDMEC cells infected with R. rickettsii Iowa strains shown in panel C were determined.Secreted IFN-β in the RARP2-expressing strain was below the limits of detection.Shown are the means ± S.D. (n = 6).Data were analyzed by one-way ANOVA using Tukey's multiple comparison adjustment.ns = not significant; * = P < 0.05; *** = P < 0.001.